Human operator errors are a common cause of failures in safety-critical systems. It is well understood that subtle flaws in user-interface design can often enable operator errors. Identifying these flaws in an interface early in the development process is an important problem. We propose a model-based technique for testing human-machine interfaces. Our modeling approach combines a model of an interactive task with a ``mistake model'' that captures the possible mistakes committed by the operator during the execution of the task. Subsequently, various combinations of mistakes defined in the mistake model are simulated along with models of the interface and the controller. Each combination is classified as potentially harmful or otherwise, depending on whether it leads to the failure of a correctness property defined in the model. Finally, the data collected from simulations is analyzed using statistical techniques: logistic regression, correlation analysis and machine learning using decision trees. The results of the analysis allow us to evaluate the impact of various mistake combinations on the success or failure of an interactive task. We present some applications of our framework to rank failures based on their effect on the outcome of the interaction. We discuss some results on the experimental evaluation of our approach over simple interfaces for drug dosage calculators commonly used in home/hospital care as well as an ongoing evaluation over a generic infusion pump (GIP) interface, developed at UPenn. Joint work with Hadjar Homaei and Clayton Lewis
Sriram Sankaranarayanan is an assistant professor in the computer science department at the University of Colorado, Boulder. His primary research interests include techniques for modeling and verification, especially of embedded and control systems. Sriram graduated with a PhD in Computer Science from Stanford University in 2005 and subsequently worked as a research staff member at NEC Laboratories America. He has been the recipient of several awards including the Siebel fellowship and recently the NSF Career award.